How big has the effect of increases in the minimum wage been on youth employment?

One of the more puzzling features of the Canadian labour market in the last few years recovery has been the stubborn refusal of youth employment rates to recover from the recession:

There may a relatively simple (partial) explanation. This is taken from a recent Statistics Canada study on trends in the minimum wage:

There seems to be a broad consensus around "small increases in the minimum wage will have small effects on employment", but this is not a small increase: the CPI-corrected Canadian minimum wage increased by 12.9% between 2008 and 2013. So in this post I'm going to try to put together some back-of-envelope numbers about what sort of effect this increase has had on youth employment.

The counterfactual I'm going to consider is: What would have happened to youth employment if Canadian real minimum wages had simply increased in line with the CPI since 2008? That is to say: what were the effects of that extra increase of 0.1215 ( = log[1.1292]) in the log of the minimum wage between 2008 and 2013?

The answer depends on what you think the relevant elasticity is. Happily, WCI colleague Livio Di Metteo has just pointed us to a recent commentary by Morley Gunderson. Prof. Gunderson has done a lot of work in this field, and his reading of the Canadian minimum wage econometric literature goes as follows:

recent evidence based on different data sets and methodologies generally finding that a 10 percent increase in the minimum wage reduces employment by about 3 to 6 percent for teens and slightly less for young adults.

In other words, the elasticity of the minimum wage on Canadian employment is between -0.3 and -0.6 for those teenagers, and (say) between -0.2 and -0.5 for those in their early 20s. He's made use of the same assessment in other studies, so that should be good enough to go on for the purposes of this post.

Let's look at teenagers first. Applying the estimates (-0.3, -0.6) to an increase of 0.1215 the log of the minimum wage results in a reduction of log employment of between 0.0364 and 0.0729. Total employment of those aged 15-19 in 2013 was 822,000, so employment would have been between 30,000 and 62,000 higher under the counterfactual of no real minimum wage growth. Here's what this means for teenage employment rates:

Teenage employment rates are about seven percentage points below the 2008 peak, and the higher minimum wage does explain some of the gap. But the explanatory power here is limited: at most 40% of the gap can be explained by the higher minimum wages.

Repeating the exercise with slightly smaller - in absolute value - elasticities in the range of (-0.2,-0.5) leads to reductions of log employment of between 0.0243 and 0.0607 for those between the ages of 20 and 24. Total employment in this group was 1.6 million in 2013, so employment would have been between 40,000 and 102,000 higher under the counterfactual of no real minimum wage growth. Here's what this means for 20-24 employment rates:

This looks quite different. The weak-elasticity counterfactual explains almost half of the 2008-2013 employment rate gap, and the strong-elasticity case generates employment rates that are higher than those observed in 2008. Of course, one of the reasons it looks different is that the drop in the employment rate for 20-24-year-olds was much less pronounced than for teenagers.

Now to put the two together. The 'lower elasticities' case is based on the sum of the two lower bounds, and the 'higher elasticities' case uses the sum of the two upper bounds.

The 15-24 employment rate in 2013 was 4.6 percentage points below was it was in 2008, and between one-third and four-fifths of this gap - equivalent to between 70,000 and 164,000 jobs - would have been closed if the rate of growth in the minimum wages had been limited to the rate of growth in the Consumer Price Index.

Comments

[SvN]

Stephen, do your calculations assume that all employment in those age brackets are at the minimum wage?
Any idea what fraction are currently at the minimum?

[Stephen Gordon]

No, these regressions look like

log[employment] = elasticity*log[min wage] + controls + e

The ratio at the minimum has - unsurprisingly - increased. It's often been remarked that the effect of the minwage on employment gets stronger as it becomes more binding: the Card-Krueger study used data from 1992, after a decade-long freeze in the *nominal* minwage under Reagan. I may get back to this point.

[Kevin Erdmann]

I'm just a blogger, and there has been tons of work on this topic - most probably much better than mine. But, I recently looked at the 6 series of minimum wage hikes over the past 60 years in the US, and I found the effect on teen employment to be similar to what you're showing with the 15-19 year olds here.
http://idiosyncraticwhisk.blogspot.com/2014/01/a-couple-more-minimum-wage-regressions.html

[L uke I]

Would it not be important to consider relative wages vis-a-vis the prime working aged population? Otherwise your counterfactual may be making some outrageous assumptions regarding the relative values between youth and prime working aged workers'wages.

[Stephen Gordon]

I don't think so. The minwage/average wage ratio also increased over this time.

[Kevin Milligan]

Morley's review is well done and he is obviously a credible source. But his survey missed a recent paper by Brochu and Green published in the EJ http://onlinelibrary.wiley.com/doi/10.1111/ecoj.12032/abstract (ungated here: http://www.economics.ubc.ca/files/2013/05/pdf_paper_david-green-impact-minimum-wages-labour.pdf).

Brochu-Green find: " The calculated minimum
wage impact implies that a 10% increase in the real minimum wage generates a 0.76% decline in
the employment rate with an associated standard error of 0.50. In comparison, for teenagers, the calculated minimum wage effect is -1.7% with a standard error of 0.9"

[Denis Drew]

I don't know about day-to-day conditions in Canada of course. But bear in mind that a higher minimum wage attracts higher value employees -- a.k.a., older employees?

Sort of the reverse of the effect of deunionizing certain jobs in the US -- like supermarket -- where the value of the new (two-tiered) employees seems lower.

In Chicago, amazingly, 100,000 out of my guess 200,000 minority, gang age males are in street gangs mostly (I presume) because among other things the US minimum wage is now $3.50 an hour below the 1968 level while per capita income has nearly doubled -- by our standards our minimum wage pays like a Bangladesh shoe factory. And incredible state I call it the "Great Wage Depression."

[Wonks Anonymous]

Denis Drew, I live in Chicago. Can you give a cite on gang membership?

[Kevin Erdmann]

Stephen, I'm confused by your response above. My analysis found that an increase in MW/AW led to a decrease in EPR. Isn't that what you're seeing here?

[Stephen Gordon]

Kevin: I was responding to the comment immediately preceding, not yours. Your results and the ones I'm citing have the same sign.

[Denis Drew]

Wonks,

http://www.cbsnews.com/news/gang-wars-at-the-root-of-chicagos-high-murder-rate/

[mpledger]

What is the "per cent" employment rate? Is it the number of people looking for work divided by them and those in work i.e. it leaves out the kids in education? When there is an economic downturn more people go into education to acquire more employable skills and to delay looking for work in a bad market. This creams the top off the pool of available workers - you have to have a plausible chance of getting a diploma to go to college - it lowers the standard of the pool of available young workers making them less employable.

The rates of young people being unemployed is rising in the Western world so that it's happening in Canada at the same time as a minimum wage rise could just be chance.

[mpledger]

When the Mayor in Chicago attempted to close a whole of schools and merged high schools, people predicted that there would be a lot of turmoil as it made kids move through neighborhoods and into schools of opposing gangs. The kids don't have much choice about which gang they belong to - they live on ? street then that sets their gang affiliation even if they never go near any member of that gang. So now gang and bystander killings are up - quelle surprise.

[Donald A. Coffin]

A similar chart for the US, where there has been no nominal minimum wage increase (and therefore a slow decline in the real minimum wage), shows and increase in the employment-population ratio for those age 20-24 (from 59.7% at its trough in January 2010 to 63.0% in June 2014). For those age 16/17 and 18/19, the corresponding changes are from 26.4% to 26.8% and from 36.3% to 38.4%. The actual increases in employment are -3.5% (16/17 year olds), +2.5% (18/19 year olds), and +1.2% (20-24 year olds) (the population base for these three age groups has been falling slightly in the US). Given a decline in the real minimum wage of 9% (the overall increase in the CPI) the (crude) employment elasticities are pretty small.

[W. Peden]

mpledger,

Youth unemployment has been rising in parts of the West, e.g. dramatically in the UK ever since the minimum wage was extended to young people. However, it's simply incorrect to say that this is a universal trend in the West. In the case of Germany, which is only now beginning to introduce minimum wages, youth unemployment rates are much as they were 20 year ago-

http://www.zerohedge.com/sites/default/files/images/user3303/imageroot/2012/05/20120502_Scary.png

- and Switzerland, which also has no minimum wage, still has extremely low youth unemployment-

http://www.quandl.com/ILOSTAT/UNE_DEAP_RT_SEX_T_AGE_AGGREGATE_Y15_24_M_CHE.png?&dataset%5Bwidth%5D=375&dataset%5Bheight%5D=250&dataset%5Bvisible_columns%5D=0&dataset%5Bgraph_title%5D=Youth%20Unemployment%20Rate%20-%20Switzerland

Of course, there are also sorts of things that could be behind these figures other than the minimum wage, and it's not clear to me that the impact of minimum wages on unemployment is well-understood even now. My point is simply that there is not a universal trend towards higher youth unemployment in the West, cyclical or structural, that can explain higher Canadian youth unemployment.

[Ralph Musgrave]

I thought youths were ALWAYS hit badly in a recession and were ALWAYS the last to find jobs come the recovery. Is Canada and the recent recession any different to other countries / recessions?

[genauer]

@ W Peden, Ralph, Stephen , .... basically everybody

Germany is different, youth unemployment just made a one year little up tick in 2009 (see the zerohedge picture above, it is now in Bavaria at 3%, just like Switzerland, one of our very few role models. One key factor to this is our apprenticeship system, we are now also exporting to Greece, Italy, Spain, USA.

The Anglo-Americans call it an un-liberal service market : - )

We are now introducing a nation minimum wage (8.5 Euro = 11 $) in Germany, kind of a heresy, but necessary, but only for above age 18, how is that in Canada?

The reason I am , as a German, extremely interested in the Canadian situation, is that for you it is this one change only, and for us this is intermixed with 4 other things:

1. demography
2. extension of free EU wide labor market in 2014 to countries like Bulgaria and Romania, with minimum wages of 1 Euro
3. a de facto minimum social income system (Hartz IV) which leads to all kinds of undesirable effects without a minimum wage
4. a deliberate attempt to stoke inflation in Germany to get to a common Euro 2.0% inflation, while others, like France, with 9.4 Euro, do some wage moderation. An effort, very unusual for us (first ever in history?), but carried out with national unity, even the holy Bundesbank came out in support.

And that means it is impossible to separate effects in the analysis of German data.

The minimum wage increase in Ontario this year, significantly above the 0.45 average wage, is the touch stone and calibration for the impact of this.

Who makes what prediction for youth unemployment end of 2014 to March 2015? Even preferable relative to the development of general employment?

[Chris J]

Stephen,

Sorry... loss of jobs in 2008 was because of minimum wage increases? There was the housing shock, the start of the great recession, the Euro crisis, our neighbour's inadequate stimulous. I don't see how a trendline dominated by a loss of jobs in 2008 is anything but the shock and poor world-wide recovery.

How did you take 2008 "out" of this analysis to leave only the minimum wage?

[genauer]

chris,

this is why we need the end of 2014 point to confirm, that "global financial crisis" was not the explanation for the unemployment rise in 2009

[Stephen Gordon]

Chris J: I didn't. Employment falls in 2009 in all three scenarios. Increasing the minimum wage during a recession worsened job losses that would have happened anywsay. Increasing the minimum wage during a recovery slows the increase in employment.

[Chris J]

Maybe. But this numerical analysis does not provide evidence of that. You would need to at least plot the trendlines of min wage workers, near minimum wage but above it (ie low wage workers whose salaries could freeze or drop in the presence of a minimum wage) and working class people. You would need to control for industry to make sure your wage slice doesn't select industries unusually harmed.

Then you could make a quantitative statement about how minwage workers fared during the recovery. And perhaps even constrain the parameters used in your model.